Light Curve Analysis of Variable Stars in the SmallMagellanic System

Future Surveys of variable stars demand new techniques not only to analyze their light curves but also to tag the variable stars according to their variability class in an automated way. We have collected a part of the largest sample of RR Lyrae stars, of which 39082 were detected toward the Large Magellanic Cloud and 6369 toward the Small Magellanic Cloud and along with these we have also collected 2574 Fundamental Cepheid. At first, we normalized the time points into phase points in the range [0, 1] by using a suitable conversion formula. We then interpolated the light curve intensities in short phase steps (0.01) within phase 0 to 1 to get 100 equi-spaced points of intensities for each light curve and then also normalized the intensities between 0 to 1 for each of the light curve both in I and V band. Once these matrices consisting of a 9270×100 arrays for I band and 8867×100 arrays for V band are prepared, we performed PCA and ICA on the same database of light curves. Our main objective of this project is to compare capabilities of Principal Component Analysis and Independent Component Analysis for classification accuracy for different classes of variable stars as it is already proved that the Principal Component Analysis method is much better than Fourier Decomposition method in order to classify the different classes of variable stars.